aramaic in iran

36
ARAMAIC IN IRAN PRODS OKTOR SKJÆRVØ INTRODUCTION This article has a dual purpose. First, I wish to remind Aramaists of a relatively important corpus of Aramaic texts on Iranian ground which has so far received but little and scattered attention. A comprehensive study of this corpus is a desideratum for both Aramaic and for Iranian studies. Second, I need to respond to a recent article by an Aramaist (Toll, “Die aramäischen Ideogramme”, 1990), in which a theory of the origin of the so-called Aramaic heterograms or ideograms (see below) in Iranian is proposed which departs from all previous theories. In my opinion the theory has a deficient material basis and therefore leads to erroneous conclusions. It is, however, the only such study by an Aramaist, and one of prominent academic lineage at that, and miscellaneous doubtful (if not wrong) forms deduced from the heterograms have now found their way into Hoftijzer-Jongeling. 1 I was encouraged by the author some time ago (letter of 9 September 1990) to (re)publish my arguments in favor of a modified “standard” theory. This is the first and best opportunity I have had to do so. IRANIAN LANGUAGES AND SCRIPTS Languages belonging to the Iranian language family were spoken in Central Asia from the 2nd millennium B.C.E. and on the Iranian plateau probably from no later than the beginning of the 1st millennium B.C.E. The first direct evidence for Iranians on the plateau comes from the Assyrian sources, in which the Parsuwas are first mentioned. For instance, on one campaign in 835 B.C.E.. Shalmaneser is said to have received tributes from 27 kings of Parsuwa. Tiglath-Pileser refers to the “mighty Medes” or the “distant Medes". ARAM, 7 (1995) 283-318 283 1 For instance, the forms in -TWN (late Sasanian period) are cited beside those in -TN, early Sasanian period, e.g., sgytwn ~ sgytn, quoted in Hoftijzer-Jongeling, vol. II, 776.

Upload: donna-hall

Post on 18-Jan-2016

79 views

Category:

Documents


6 download

DESCRIPTION

xc

TRANSCRIPT

Page 1: Aramaic in Iran

ARAMAIC IN IRAN

PRODS OKTOR SKJÆRVØ

INTRODUCTION

This article has a dual purpose. First, I wish to remind Aramaists of arelatively important corpus of Aramaic texts on Iranian ground which has sofar received but little and scattered attention. A comprehensive study of thiscorpus is a desideratum for both Aramaic and for Iranian studies. Second, Ineed to respond to a recent article by an Aramaist (Toll, “Die aramäischenIdeogramme”, 1990), in which a theory of the origin of the so-called Aramaicheterograms or ideograms (see below) in Iranian is proposed which departsfrom all previous theories. In my opinion the theory has a deficient materialbasis and therefore leads to erroneous conclusions. It is, however, the onlysuch study by an Aramaist, and one of prominent academic lineage at that, andmiscellaneous doubtful (if not wrong) forms deduced from the heterogramshave now found their way into Hoftijzer-Jongeling.1 I was encouraged by theauthor some time ago (letter of 9 September 1990) to (re)publish my argumentsin favor of a modified “standard” theory. This is the first and best opportunityI have had to do so.

IRANIAN LANGUAGES AND SCRIPTS

Languages belonging to the Iranian language family were spoken in CentralAsia from the 2nd millennium B.C.E. and on the Iranian plateau probably fromno later than the beginning of the 1st millennium B.C.E. The first direct evidencefor Iranians on the plateau comes from the Assyrian sources, in which theParsuwas are first mentioned. For instance, on one campaign in 835 B.C.E..Shalmaneser is said to have received tributes from 27 kings of Parsuwa.Tiglath-Pileser refers to the “mighty Medes” or the “distant Medes".

ARAM, 7 (1995) 283-318 283

1 For instance, the forms in -TWN (late Sasanian period) are cited beside those in -TN,early Sasanian period, e.g., sgytwn ~ sgytn, quoted in Hoftijzer-Jongeling, vol. II, 776.

Page 2: Aramaic in Iran

His campaigns against them took him as far as Mount Bikni, which is probablyto be identified with Mount Alvand, south of modern Hamadan (ancient Ec-batana). Finally, at the battle of Halule on the Tigris in 691, the Assyrianking Sennacherib (704-681 B.C.E.) faced an army of troops from Elam,Parsumas, Anzan, and others, and in the Vassal Treaties of Esarhaddon(680-669 B.C.E.) and elsewhere numerous “kings” of the Medes are men-tioned.

The earliest direct evidence for Iranian languages is the corpus of Old Persianinscriptions from the Achaemenid period, the first of which was probably thegreat inscription of Darius I at Bisotun dating from 520-519 B.C.E. and for thewriting of which a cuneiform alphabet was invented by the king’s scribes athis order.

Old Persian (the ancestor of Middle and modern Persian) was apparentlynot used as an administrative language, however, which remained Elamite inthe royal administration of Persepolis and Susa, written on permanent mater-ial, and Aramaic presumably for letters and other documents, written mostlyon perishable materials. It was therefore during the Achaemenid periodthat Aramaic started spreading throughout the Iranian territories as scribal lan-guage and the Aramaic script became the primary means of writing. TheAchaemenid satraps of Asia Minor inscribed their coins using Aramaic, andso did the Seleucid (after the death of Alexander in 323 B.C.E., with an inter-lude of Greek) and Parthian kings. It reached Bactria (Afghanistan) no laterthan the 2nd century B.C.E., when we find it used for Iranian translations ofAsoka’s rock edicts. In these inscriptions it is also, exceptionally, used forwriting Middle-Indic languages (Prakrit).

The successors of the Seleucids, the Parthians (Arsacids) used Aramaicscript for all writing: coins, letters, inscriptions, etc., and a typical Parthianductus appears in royal inscriptions from the 2nd century C.E.

Further away in Central Asia the Aramaic script was used for writing Sogdian(in the area of modern Uzbekistan and Tajikistan) no later than the 3rd centuryC.E., as well as for Chorasmian (in the area of modern Turkmenistan). TheSogdian variant of the Aramaic script, the earliest version of which is seenin some letters dating from the 3rd century C.E., later developed into severalcursive variants, referred to as the Sogdian and (most cursive) Uigur scripts,as it was also used to write Old Turkish.

The farthest extension of the Aramaic script was into the Tarim Basin,modern Chinese Turkestan or Xinjiang, where, in the first centuries of ourera, it formed the basis for the development of the KharoÒ†hi script used towrite the local Middle-Indic (Prakrit) language. The innovation of the

284 ARAMAIC IN IRAN

Page 3: Aramaic in Iran

KharoÒ†hi script was to express vowels by modification of the basic letter,e.g., by subscripts, a method adopted from the Indian scripts.

The only Iranian languages that did not, apparently, use the Aramaic script,were Bactrian and Khotanese and its relatives. Bactrian was the languagespoken in Bactria, which was settled by Alexander’s soldiers in the 3rd centuryB.C.E., who introduced the use of Greek script. Whether the Bactrians usedAramaic before this time is not known.

Khotanese and related languages were spoken in the Tarim Basin,in Khotan on the southern Silk Route and in Kucha and Agni on thenorthern Silk Route. The populations of these areas early on became Bud-dhist and adopted variants of the Indic Brahmi script for writing theirscriptures.

On the Iranian plateau the Aramaic script continued to be used in theSasanian period for writing Parthian and Middle Persian, the descendant ofOld Persian and ancestor of modern Persian. The earliest monuments,from the 2nd-3rd century C.E., are in a lapidary ductus, but, parallel to it,there must have existed ductuses more adapted to writing on parchmentand papyrus and, later, paper. We may note that the transition from the“classical” Aramaic ductus to the Middle Persian ductus took place ca.300 C.E., as can be seen from the coins. The earliest non-lapidary ductus isseen in a manuscript containing a part of the Psalms of David (the PahlaviPsalter) discovered in Chinese Turkestan. This ductus soon developed intoa highly cursive one, the Book Pahlavi script, which is the standard scriptused in the literature of the Zoroastrians, as well as on seals and coinsfrom the later Sasanian period. The latest and most cursive variant of thisscript is seen on papyri and parchments from the end of the Sasanianperiod. As an official script, Pahlavi was by then replaced by Arabic, how-ever.

It was also in the Sasanian period (ca. 500 C.E.?) that an alphabet based onthe Psalter and Book Pahlavi scripts was invented for the writing down of theAvesta, the ancient holy scriptures of the Zoroastrians, composed in Avestan,an Old Iranian language spoken in two different chronological stages, Old andYoung Avestan, in Central Asia and (north)eastern Iran around the middleof the 2nd millennium and in the first half of the 1st millennium B.C.E.,respectively. The inventors of the Avestan alphabet combined the forms ofthe letters (ultimately derived from Aramaic) with the principle of phoneticspelling of the Greek script, adding vowels, to produce a phonetically exactscript, by which one could record the minutest phonetic details of the liturgicalpronunciation of the holy texts.

PRODS OKTOR SKJÆRVO 285

Page 4: Aramaic in Iran

From the early Sasanian period on two varieties of Syriac script were usedto write Iranian languages, as well: the Manichean script, a variant ofEstrangelo the invention of which is ascribed to Mani, the founder ofManicheism himself, and the Nestorian script. The Manichean script wasespecially adapted to Iranian needs and was used to write Parthian, MiddlePersian, Bactrian, Sogdian, and even Tokharian (a non-Iranian Indo-Europeanlanguage spoken in several dialects on the northern Silk Route in the areas ofKucha and Agni/Qarashahr) and Old Turkish. The Nestorian script was usedin Chinese Turkestan by the Sogdian Christians.

In the early Islamic period, Hebrew was used extensively by the Jewishpopulation of Iran to write mostly modern Persian (Judeo-Persian), but alsolocal dialects, for instance, that of Hamadan (ancient Ecbatana).

Among modern Iranian languages Persian has a literature written in Arabicscript reaching back to the beginnning of the Islamic period. Other literarylanguages, such as Pashto, Kurdish, and Baluchi, also use (or have used) theArabic alphabet, with numerous modifications.

For a survey of the Iranian variants of the Aramaic and Syriac scripts seeSkjærvø, “Iranian Alphabets".

ARAMAIC AND ARAMEO-IRANIAN TEXTS FROM IRAN AND THE USE OF ARAMAIC

HETEROGRAMS

The earliest extant Aramaic texts written by Iranians, or scribes in the serviceof Iranians, are the Aramaic texts from the Achaemenid period. These comprisethe Aramaic texts from Persepolis from the early 5th(?) century B.C.E., mostlyshort inscriptions on ritual utensils,2 and the Aramaic version of the Bisotuninscription found among the papyri from Elephantine (5th century). TheAramaic texts from Egypt are also heavily influenced by Old Persian officialand religious terminology.

An Aramaic inscription on the tomb of Darius at Naqs-e Rostam may befrom the Seleucid period (3rd-2nd centuries B.C.E.), if Henning’s reading of thename slwk in it is correct (“Mitteliranisch”, 24). Unfortunately, Henning did notspecify where in the inscription he saw this name, and nobody has seen it since.No other Aramaic text from Seleucid Iran other than on coins survives.

From the 2nd-1st centuries B.C.E. we have the Aramaic versions of the rockedicts of Asoka found in Afghanistan. The Aramaic versions are accompaniedby Greek versions or Indic versions in Aramaic script.

286 ARAMAIC IN IRAN

2 Ed. Bowman, Aramaic Ritual Texts; among reviews, see Levine.

Page 5: Aramaic in Iran

From the Parthian period on the question arises whether the texts we haveare Aramaic or Iranian written with Aramaic ideograms/heterograms (see below).The documents in question are a land sale document found at Awroman inmodern Kurdistan and potsherds inscribed with wine receipts from Nisa, thenorthern capital of the Parthian empire, both dating from the 1st cent. B.C.E.,and a couple of Parthian royal inscriptions and some inscriptions found innorthwestern Iran, all from the 1st to 3rd cents. C.E.3

The following examples of texts are divided into three groups: The firstgroups contains texts that are indisputably in Aramaic language and the thirdtexts that are as clearly in Iranian languages (Middle Persian and Parthian). Inthe texts of this third group we see numerous Aramaic words, in increasinglyless “correct” form as time goes on. From several circumstances it has longbeen known that these Aramaic words were merely a scribal devise to write thecorresponding Iranian words (see, e.g., Westergaard, Zendavesta, 20 fn. 2; Sale-mann, “Mittelpersisch”, 250; Schaeder, Iranische Beiträge, 206-209). On onehand, in different copies of the same text, the Aramaic words vary with Iranianwords (e.g., Mid. Pers. bay “god(s)” is sometimes written “phonetically” asbgy, sometimes “heterographically” as ORHYA, literally ¨rÌyˆ < Aram. ˆlhyˆ“the gods”). On the other hand, the Aramaic words frequently receive “end-ings” that properly belong to the Iranian words (e.g., BRE < b¢reh, literally “hisson", written for pus “son", but BRE-r written for pusar “of the son, sons”with Persian ending -ar). For this reason these Aramaic words are now com-monly referred to as “heterograms” or (increasingly less commonly) “ideo-grams". This kind of scribal practice is, of course, well known from Me-sopotamia, although a historical connection between the two is uncertain.4

The second group contains texts from the early-mid Parthian period. Thenature of the language of these texts is debated. Some regard it as heterographicIranian (Parthian) others as Aramaic, although written in faulty orthography.The main argument in favor of the first hypothesis is the faulty orthography andthe occurrence of Parthian words. Against this hypothesis and in favor of thesecond is the fact that the Aramaic elements in these inscriptions do not quiteconform to the heterographic system of the later Parthian inscriptions. My owninclination is to regard these texts as written in the kind of unskilled Aramaicthat was soon to give way to heterographic Iranian (see also below).

PRODS OKTOR SKJÆRVO 287

3 Among Aramaic texts written by Aramaic-speaking communities in Sasanian Iran, wemay mention the Palmyran texts.

4 All the languages written in the “classical” Aramaic script use heterograms, thoughonly Parthian and Middle Persian to a greater extent. Texts in the Syriac “Manichean” scripthave no heterograms.

Page 6: Aramaic in Iran

It is customary to transliterate heterograms using capital letters (roman oritalic), but there are several current systems. I will be using that first employedby D. N. MacKenzie, who dispenses with diacritical signs by assigning thecapital letters A, E, O to ˆalep, he, and ¨ayin, leaving H for Ìe†; C is used forÒa∂e and Q for †e†. This system has the advantage of dissociating the hetero-graphic spellings from the original Aramaic ones. Gignoux’s Glossaire fol-lows a more traditional system, using ˆ, ¨, H, Î, and ™ (but C, not ∑), for in-stance: ¨RÎYˆ (see above), ÎTYˆ/Parthian ΙYˆ “arrow” (< Aram. ̆yˆ “thearrows”) and YCBH “to wish” (< Aram. yÒbh “he wishes”), as opposed toORHYA, HTYA/Parthian HQYA, and YCBE, used here.

Note also that, while the Parthian heterograms are faithful to the Aramaicorthography, in Middle Persian the letters Q and Q (™) have been replaced byK and T5; example: Parthian QQL “kill", Middle Persian YKTLWN. Notinfrequently we find O for A, as in ORHYA, Parthian ALHA (< Aram. ˆlhˆ).H (< Ì) is normally used in non-final position and E (< h) in final position foreither Ì or h in both Parthian and Middle Persian; example: Middle PersianYKOYMWN, cf. Parthian HQAYM-, literally Ìqˆym- for hqym(?).

I. ARAMAIC TEXTS

Aramaic texts from Persepolis

Most of the Persepolis inscriptions published by Bowman are of the fol-lowing type:

bprkn byrtˆ lyd NN1 sgnˆ (rbˆ) In the parkan (precinct?) of the fortress, for NN1,the (grand) segan,

NN2 ¨bd/¨bdw ˆbswn6 znh NN2 made (is the maker of?) this abson (pestle).

The language of these texts is clearly Aramaic, not Old Persian in Ara-maic disguise. In Old Persian the genitive normally precedes the noun itqualifies, as does the demonstrative pronoun; cf. the similar Old Persianinscriptions from A3Pa: imam ustasanam aqaganam mam upa mam kærta,literally: “this staircase of stone by me under me (= during my reign) wasmade".

288 ARAMAIC IN IRAN

5 Exceptions: Middle Persian QB, later TB new “good”; QDM written MDM (with Q = M)abar “on”, cf. Parthian QDMTE parwan “before”.

6 Iranian words in the text are in italics here.

Page 7: Aramaic in Iran

The form ¨bdw I assume contains the personal pronoun -(h)u, added eitherto the perfect or to the present participle: *¨¢ßa∂-u7 or ¨aße∂-u.8

An Aramaic version of the edict of Asoka: Kandahar 1

The following Aramaic inscription is accompanied by a Greek version;see the latest edition by Pugliese Carratelli and Garbini. The awkward syntaxis as likely to have been caused by the Indic original as by the influence of anIranian scribe.

1 snn 10 ptytw ¨byd zy mrˆn (For?) 10 years expiation (has been) made (or:prydrs mlkˆ qsy†ˆ mhqs† is making?) (he) who (is) Our Lord, Priyadarsa,

the king, the promoter of truth,2 mn ˆdyn z¨yr mr¨ˆ lklhm ˆnsn Since then evil (is) less for all people, and all

wklhm ˆdwsyˆ hwbd hostilities he has eliminated.3 wbkl ˆrqˆ rˆm sty wˆp zy znh And in all the earth (is) peace and happiness.

bmˆklˆ lmrˆn mlkˆ z¨yr And *in addition, for eating (= food) for OurLord the king (there are) less (people)

4 q†ln znh lmÌzh klhm ˆnsn who kill. This is for all people to see. They have ˆthÌsynn wzy nwnyˆ ˆÌdn held themselves back also (those) who catch fish

5 ˆlk ˆnsn ptyzbt knm zy prbst those people declared (against it?); similarly,hwyn ˆlk ˆthÌsynn mn those who were *trappers, those have held them-

selves back from6 prbsty whwptysty lˆmwhy * trapping. And (they are) obedient to his mother

wlˆbwhy wlmzystyˆ ˆnsn and to his father and the elders, people (are),7 ˆyk ˆsrhy Ìlqwtˆ wlˆ ˆyty dynˆ as destiny has laid it down. And there is no

klhm ˆnsyˆ Ìsyn judgment for all men pious.8 znh hwtyr lklhm ˆnsn wˆwsp This has benefited all men and over and over

yhwtr will benefit (them).

The uncertainties about the exact meaning of the Iranian word ptytw andthe function of zy make it impossible to be certain about the forms of theverbs in the first line; if ¨byd and mhqs† are parallel, then ¨byd will be ¨aße∂(cf. ˆthÌsynn = e†haÌsen- and hwtyr = hawter). The 3rd plur. forms in -n agreewith ˆnsn “people” and are presumably for -in (Segert, AltaramäischeGrammatik, 184 §5.2.3.4.3).

PRODS OKTOR SKJÆRVO 289

7 Cf. the enclitic emphatic (h)u in Syriac, Nöldeke, Syrische Grammatik, 167 §221? It isnot clear from the handbooks to what extent this was a “genuine” Aramaic practice; it maybe (if my interpretation is correct) just an Iranian practice.

8 Differently Bowman, Aramaic Ritual Texts, 40.

Page 8: Aramaic in Iran

II. ARAMAIC OR IRANIAN (PARTHIAN)?

The land sales document from Awroman in Kurdistan

1 snt 300 yr̈ ˆrwtt mzbnw ptspk Year 300, month (H)arwatat. The seller is Ptspk,bry tyryn son of Tiren, who (is) from Brkn, (of) the vine-

2 zy mn brkn (?) krmˆ ˆsmk mh yard (of) Asmak, which is half a share (yat) ofˆbykskn plg yˆt 9 Abikasakan.

3 wzbnw ˆwyl bry bsnyn kzy ˆÌy And the buyer is Awyl, son of Bsnyn, as long asklˆ zwzn 20 20 20 11111 I live(?),11 for a total of 65 drahms,

4 mh mn bwmÌwtw (?) ˆt. Ì(r)w10 which… from the *land-lord (bum-xwataw).Ìmy ˆklw qdmth These *swore before him

5-6 sÌdyn [names] (as) witnesses: [names]7 […..] krmˆ ˆsmkn krmˆ zbnt ˆwyl […] the vineyard (of) Asmakan (!). The vine-

mn yard, I Awyl have bought from8 ptspk klˆ zwzn 20 20 20 11111 Ptspk for a total of 65 drahms.

On this text see Henning, “Mitteliranisch”, 28-30. Only the orthographywould seem to speak against this document being Aramaic, such as Ì for h,while the position of the verbs at the beginning of clauses, although found inthe Parthian and Middle Persian inscriptions, would be unusual(?) in a Parthiandocument. The use of the verb “eat” for “swear” has been recognized as anIranian calque: sogand xwardan “eat > swear an oath".

Nisa, Wine receipt, document no. 411

See Diakonoff and Livshits, Parthian Economic Documents, 43.

1 bÌwtˆ znh mn krmˆ In this jar from the vineyard 2 ˆwzbry zyb brzmytn uzbariy which (is) in Brzmytn3 nwk qry Ìm 10 1111 1111 1 new (wine) *called, 19 mari. 4 hn¨lt ¨lsnt 200 1111 111 I (?) [as producer] delivered (it) for (?) the year

2075 hyty ˆwgtnwk Ugtanuk brought (it) 6 mdwbr zymn the wine-deliverer who (is) from7 brzmytn Brzmytn.

290 ARAMAIC IN IRAN

9 Cf. Cowley (Aramaic Papyri, 1-2) Papyrus no. 1 lines 2-3 yhbn lky plg mn[t]ˆ zy yhbwln… “we have given to you half the share which was granted to us…”

10 Letter following t small letter, second last letter d, r, k. Either Aramaic or Iranian. 11 Cf. kdy Ìyˆ “as long as (there is) life”, Hoftijzer-Jongeling, Dictionary, Vol. I, 317, 4-

5th lines from bottom.

Page 9: Aramaic in Iran

The use of the masc. form znh with a fem. noun in bÌwtˆ znh is un-Aramaic.12 The combination of a preposition plus a noun, as in ¨lsnt, is nototherwise found among the known Parthian and Middle Persian heterograms.The forms and meanings of hn¨l-t (beside hn¨l-w) as opposed to hyty are notclear; hn¨l-t may be a heterogram, as suggested by other pairs such asYNTN-t ~ YNTNW. On the other hand, the Nisa documents do not yet showthe confusion of h and Ì typical of Iranian texts.

The inscription from McÌeta (Armazi)

On this text see Altheim and Stiehl, Supplementum, 74-85

1 ˆnh s¨rpy† brty zy I, Serapeitis, daughter of2 zwyÌ qlyl b†Ìs zy prsmn Zeouakhes the younger, bitaxs of Porasman 3 mlkˆ ˆntt zy ywdmngn wnÒyÌ the King, wife of Yodmangan. And he was vic-

torious4 wkbyr ˆrwst ¨byd(w)ˆ rb and performed great deeds, master of5 trbÒ zy hsyprnwg mlk bry court (procedures) of King Ksefarnoug, son6 zy ˆgryp rb trbÒ zy of Agrippa, master of court (procedures) of 7 prsmn mlk Ìbl Ìblyk mˆ King Porasman. Woe upon thee! that 8 zy prnws lˆ gmyr whkyn the full age (*p¢rnaˆus) was not completed. And

so9 †b wspyr yhwh hyk zy br good and beautiful is she that a son

10 ˆyns lˆ dm¨ yhwh mn of men is no equal with respect to11 †bwtˆ wmˆytyn bsnt 10 10 1 goodness. And I(?) died in (my) year 21.

Notes: On nÒyÌ … ¨byd(w)ˆ as naÒeÌ… ¨aße∂-(h)wa see Altheim andStiehl (Aramäische Sprache, 303). As for mˆytyn = *maˆe†-en “I am dying”(cf. myytnˆ Dalman, §65, p. 289), which as suggested by Altheim and Stiehl(ibid., 47) has masc. instead of fem. participle, compare the same practice inImperial Aramaic from Asia Minor and Egypt according to Segert (Alt-aramäische Grammatik, 330-332 §6.3.1.3.56).

III. IRANIAN INSCRIPTIONS

Inscriptions of the standard 3rd-century type are known from the 2nd centuryon. In these heterograms are common, but restricted to a relatively limited setof words and forms. The exact pronunciation of the Iranian words can bededuced from the Manichean texts and etymology.

PRODS OKTOR SKJÆRVO 291

12 See Henning, “Mitteliranisch”, 27-28.

Page 10: Aramaic in Iran

Parthian royal inscriptions

• Reign of Arsak Walgas, son of Mihrdad (Miqrdat), 151 C.E., see Morano,“Contributi",, and Skjærvø, review of Gnoli and Panaino, (eds.), and reviewof Skalmowski and Tongerloo, (eds.).

1 … ˆrsk wlgsy MLKYN MLKA13 [In the year…] Arsaces Vologases, King of Kings,… Arsak Walgasi sahan sah

2 BRY mtrdt ML[KA] son of King Mitradates,puhr Mihrdad sah

3 [KT]SW OL mysn BRA mtrdt fought in Mesene against King Mitradates, son ofMLKA BRY*kosed o Mesan *abar Mihrdadsah puhr

4 pkwr MLKYN MLKA Pacorus, King of Kings.Pakor sahan sah

5 mtrdt MLKA MN TME MRDPW He chased King Mitradates from there,Mihrdad sah az od …

6 hmk mysn AHDW he took all of Mesene.hamag Mesan girwed

7 ZNE ptkr wrtrgn ALHA14 ME This image of the god Warhagn (Herakles), whichMN mysn HYT-t15 was brought from Mesene,im padkar Warh(r)agn bag ceaz Mesan awurd

8 nygndn B tyry bgny HQAYMW he placed as *trophy in the temple of Tir. *nigandan andar Tiribaginiawested

• Reign of Artaban, 215 C.E. (see Henning, “Mitteliranisch”, 40 41; Altheimand Stiehl, Supplementum, 98).

1 SNT 400 20 20 20 11 YRHA Year 462, month Spandarmat, day Mihr,spndrmty YWMA mtry16

sard 462 mah Spandarmad rozmihr

292 ARAMAIC IN IRAN

13 Thus also in the later Parthian inscriptions, from Aramaic mlkyn mlkˆ. Middle Persian,however, uses MLKA-n MLKA, with -n as phonetic complement for the ending -an.

14 Parthian ALHA < ˆlhˆ = bag, Middle Persian ORHYA = bay, literally means “god”,but is mostly found in the plural in the sense of “majesty”.

15 Parthian HYT-t is an alternate spelling for HYTY-t. On the final -t in this type ofheterogram (also BNY-t in the next inscription, see the discussion below).

16 The dating formulas differ in all the languages using heterograms: Parthian SNT…YRHA… YWM… = Middle Persian SNT… BYRH… YWM… = Sogdian SNT… YRHA…YWMA… = Choresmian BSNT… YRHA… BYWM….

Page 11: Aramaic in Iran

2 ˆrtbnw MLKYN MLKA (reign of?) Artabanos, king of kings, Ardban sahan sah

3 BRY wlgsy MLKYN MLKA son of Vologeses, king of kings.puhr Walgas sahan sah

4 BNY-t cy[t]k ZNE LY hwsk *Husak, satrap of Susa, built this stele.17

sws hstrp*dist cidag im man HusagSus sahrab

Sasanian royal inscriptions

• Reign of Ardaxsahr I (224-239/40; see Herzfeld, Paikuli, 85).

Parthian ptkr ZNE mzdyzn [ALH]A ˆrthstr MLKYN MLKA ˆryˆn MNW[syh]r MN yˆztn padkar im mazdezn bag Ardaxsahr sahan sah aryan ke sihr azyazdan

Middle Persian ptkly ZNE mzdysn bgy ˆrthstr MLKAn MLKA ˆyrˆn MNWctry MN yztˆn pahikar en mazdesn bay Ardaxsahr sahan sah Eran ud an-Eranke cihr az yazdan“This picture is of the Mazdayasnian Lord Ardaxsahr, King ofKings, whose seed is from the gods”

Parthian BRY ALHA pˆp[k] [M]LKApuhr bag Pabag sah

Middle Persian BRE bgy pˆpky MLKApus bay Pabag sah“son of the Lord King Pabag".

• From the inscription of Sapur (Sabuhr) I at Naqs-e Rostam:

Parthian W LH-w18 ME ZNH-n19 ˆtrwn YNTN-t20 W ME ˆbdyn HQAYMW-tud ho ce imin aduran dad ud ce abden awestad

PRODS OKTOR SKJÆRVO 293

17 For this type of “colophon”, cf. also Middle Persian nibist Boxtag dibir “Boxtag thescribe wrote (this inscription)” (inscription of Kerdir at Naqs-e Rajab).

18 Parthian LH-w = ho can also be read as LHW < lehu (less likely as L-hw).19 Parthian ZNH-n: note that ZNE (< znh) before the phonetic complement of the plural

ending becomes ZNH- (literally znÌ-) according to the rule that E is used only in final position(thus also LH-, literally lÌ-< lh). Note that in this inscription ZNH-n imin “these” correspondsto Middle Persian LZNE-sn = imesan, whereas in the next inscription Parthian ZNE = im cor-responds to Middle Persian ZNE = en “this”. See toward the end of this article on thepronominal heterograms.

20 On the final -t of Parthian YNTN-t and HQAYMW-t see below.

Page 12: Aramaic in Iran

Middle Persian W ZK ZY LZNE-sn ˆtwrˆn YHBWN W ZY-sn PWN ˆdwynHNHTWNud an i imesan aduran dad ud i-san pad ewen nihad“and that which to these fires (was) given and which for them(was) as custom established”

• From the inscription of Sapur (Sabuhr) I at Hajjiabad:

See Nyberg, Manual I, 122-123. Pa W AMT LN ZNE HQYA SDY-t QDMTE hstrdryn BRBYTA-n RBA-n W

ˆzˆtn SDY-tud kad amah im tigr wist parwan sahrdaran wispuhran wazurgan ud azadan wist

MP AP-n21 AMT ZNE HTYA SDYTN ADYN-n LOYNY strdrˆn W BRBYTA-n Wwclkˆn W ˆzˆtn SDYTNu-n ka en tir wist eg-in pes sahriyaran ud wispuhran ud wazurgan ud azadanwist“And when this arrow (We) shot, then before the landholders, princes, grandees,and nobles (We) shot”

Pa NGRYN22 pty ZK wym HQAYMW-t W HQYA LCD LH-w syty LBRA RMY-tpad pad ed wem awestad ud tigr tar ho cid o beh abgand

MP AP-n LGLE PWN ZNE drky HNHTWN AP-n HTYA LCD-r ZK cytˆk BLALMYTNu-n pay pad en darrag nihad u-n tir tar an cidag be abgand“and foot on this stone/crack (We) placed, and (We) the arrow beyond thatcairn away (We) shot".

DEVELOPMENT OF THE VERBAL HETEROGRAMS

The best description of the origin of the use of heterograms to write Iran-ian languages is, in my opinion, that of Henning (“Mitteliranisch”, 31-32),which deserves to be quoted in full (my translation):

Already in Achaemenid times it had become the habit to sprinkle theAramaic text with Iranian words, at first, titles, technical terms, and thelike. In this way they learnt how to write indigenous words with Aramaicletters. In the course of time, the number of Iranian words grew, at the same

294 ARAMAIC IN IRAN

21 In AP-n… ADYN-n… AP-n… AP-n, -n is the enclitic agential pronoun 1st plur. (ofmajesty) attached to the conjunction/particle AP- u(d) “and”, not a phonetic complement.Note that only in the Middle Persian is the agent expressed repeatedly, while in the cor-responding Parthian sentences the tonic pronoun LN “We, Us” is used once and is notrepeated. On the agential construction of the transitive “simple past” see below.

22 Note Parthian NGRYN < *ngryn < riglayin (dual) = Middle Persian LGLE < rigleh.

Page 13: Aramaic in Iran

time that Aramaic was increasingly neglected: It must have been difficult al-ready in the 3rd century B.C.E. to find enough trained people to write it. Grad-ually, we must assume, the word order yielded to that of the scribe’s own lan-guage, while the individual words maintained their Aramaic inflectionalforms. When this stage was reached, it is hard to tell whether we should callthe language bad Aramaic or heterographic Iranian. Subsequently, the Ara-maic inflections are gradually given up and are only used in a few fixed forms.

The idea of fixed Aramaic inflectional forms or “frozen” forms, as I shallrefer to them here, was also adopted by Nyberg (e.g., Manual, II, 1) andRosenthal (“Aramaic”, 256).

According to Henning (“Mitteliranisch”, 25) heterograms are first attestedon a coin dating from the end of the 2nd century B.C.E., where we have theform BRE, literally “his son” for regular BR “son of". Whether this isolatedexample proves that an official heterographic system of writing had beenestablished by this time, I doubt, as the 1st-century B.C.E. texts still appear tohave more Aramaic in them than the later ones. I think perhaps the final system,the one we see in the earliest Sasanian inscriptions, was established in the1st-2nd centuries C.E., when the typical ductuses of Parthian and Middle Persianwere introduced.

My own research in this matter stems from my work on the Old Persianverbal system (Undersøkelser, 1974). Here I for the first time remarked thatthe Middle Persian heterogram OBYDWN- only represents Middle Persianpresent tense forms. Later I noticed that a similar conclusion had alreadybeen reached by Herzfeld (Paikuli, 57). I continued investigating the useof the verbal heterograms in the Sasanian inscriptions (Parthian and MiddlePersian) and outlined my preliminary results in Humbach and Skjærvø, Paikuli,Part 3.2 (1983) together with a syntactic analysis of the Parthian and MiddlePersian case system published in “Case in Inscriptional Middle Persian”(1983). I elaborated and refined the “frozen forms” theory in “Verbs inParthian” (1986) and presented a detailed study of the use of the verbalheterograms in “Verbal Ideograms” (1989). Following is a summary of themain results of these studies.

The verbal systems of Parthian and Middle Persian were based on afundamental opposition between (1) present-imperfect: forms from the presentstem with a narrative-descriptive function used for the present and imperfecttenses and the moods of the present tense (indicative, subjunctive, optative,imperative)23 and (2) simple past: forms from the past stem (historically =

PRODS OKTOR SKJÆRVO 295

23 The future was expressed by the present indicative or subjunctive.

Page 14: Aramaic in Iran

past participle), originally used to denote the result of a past action, process,or state, but in the course of time, as the imperfect was lost, became the generalpast tense. Examples:

Present Imperfect Simple past

sawem “I go” *sawen “I went” sud hem (lit. “I am gone” > “I have gone,I went”)

nibesem “I write” *nibesen “I wrote” u-m nibist (lit. “by me written” > “I havewritten, I wrote”)24

The heterograms as we see them in the Parthian and earliest MiddlePersian text corpus reflect this fundamental division as follows:

In Parthian, heterograms denoting forms of the present stem end in -Wor -E, while those denoting forms from the past stem end in -T/-t25 (on whichsee below), e.g.,

“take” “come” “write” “give” “bring” “place”

Present AHD-W ATY-E ? YNTN-W HYTY-W HQAYM-Wgirw- as- dah- awar- ist-

Simple past AHD-T/t ATY-T/t KTYB-t YNTN-t HYT(Y)-T/t HQAYM(-W)-tgrift agad nibist dad awurd istad

In Middle Persian, the verbal heterograms are used for forms from bothstems, except that some verbs reserve the heterograms for forms from thepresent stem only, while the past tense is written “phonetically",26 e.g.:

“take” “kill” “do, make”

Present OHDWN YKTLWN OBYDWNgir- ozan- kun-

Simple past OHDWN YKTLWN kltygrift ozad kerd

Seeing that many of the heterograms, at least superficially, were formallyimperfects in Y-, participles in M-, and, apparently, perfects, I made theassumption that the heterograms originally corresponded to the function ofthe Iranian form they represented, that is, an Iranian present stem form wouldbe represented by an imperfect or an active participle, and a past stem form

296 ARAMAIC IN IRAN

24 Like Syriac k†iß-li. 25 Thus already Herzfeld, Paikuli, 54. 26 Thus already Herzfeld, Paikuli, 57.

Page 15: Aramaic in Iran

would be represented by a perfect or a passive participle. Thus, I proceededto assign each of the actually attested types of heterograms to one of thesecategories.

This procedure led to the realization that forms such as Parthian OBD-Wand Middle Persian OBYD-WN, which were restricted to representing presenttense forms, most likely are from active present participles with the additionof the “enclitic” subject 3 sing. personal pronoun -(h)u: ¨aße∂-(h)u, as firstsuggested by Altheim and Stiehl (Aramäische Sprache, 303). This impliestwo assumptions: (1) plene spelling of short i/e and (2) spelling of the en-clitic 3rd sing. masc. pronoun without h as -w.

(1) The plene spelling of the short i/e in the present participle is well knownin Aramaic (Dalman, Grammatik, 284-285), although it is not registered forImperial Aramaic in Segert (Altaramäische Grammatik, 266). Nevertheless,there are several instances in the Aramaic texts from Iran in which short i/eis written plene.

The earliest example may be the form ¨byd, found twice in the AramaicBisotun inscription; in both instances the editors assume it is the past participle¨¢ßi∂ “done” (Greenfield and Porten, The Bisutun Inscription, 31, 47). Suchan interpretation is not without problems, however. In line 66 we read […]hwd¨ ˆyk zy ¨byd ˆnt wˆyk hlktk, which corresponds to Old Persian azda kusuvaciyakaram ahi “make known of what sort you are!” Thus, the Aramaic ex-pression ˆyk zy ¨byd ˆnt corresponds to OPers. ciyakaram ahi, which literallymeans “what-doing you are". Sims-Williams (“The Final Paragraph”, 4) fol-lowed by Greenfield and Porten (The Bisutun Inscription, 47) suggested that¨byd ˆnt is the Aramaic rendering of the Old Persian “ergative” constructionseen in mana kærtam “I have done", literally, “my/by me done".27 Thisphrase, however, is expressed in various Aramaic dialects as ¨¢ßi∂ li, not¨¢ßi∂ *ana, which could only mean “I (was) made". It is therefore muchmore probable that ¨byd ˆnt is for ¨aße∂ ant “you are doing". In line 17 weread ˆÌr ddrs mn[d¨m l]ˆ ¨byd mkt[r ly…] “Afterward, Dadarsi did nothing(but) waited for me…” Here, as well, the interpretation as present participleis preferable in view of the following participle mktr, which can hardly beother than a present participle itself.

(2) As for the assumed spelling of the enclitic 3rd sing. masc. pronounwithout h, this is, as far as I can see, not found in standard Aramaic anywhere

PRODS OKTOR SKJÆRVO 297

27 Note that this is not a case of the passive participle being used in an active sense, asSims-Williams and Greenfield and Porten have it; it is not the participle that has active sense,it is the whole construction that has active function.

Page 16: Aramaic in Iran

(except in mnw “who”); in most Aramaic dialects the h of the enclitic 3rdsing. pronouns -hu and -hi had become silent from the beginning of the literaryera, however, so a phonetic spelling can easily be imagined. Here again theearliest example is, perhaps, found in the Achaemenid Persepolis texts (seeabove), in which ¨bd alternates with ¨bdw without, apparently, any change toplural subject. From the Parthian period, the forms zbnw and mzbnw of theAwroman document (see above) are most easily explained as present participles+ the enclitic pronoun: zaßen-u and m¢zabben-u.

In addition I assumed that Parthian OBD-T and HZY-T were actually theold 1st sing. perfect = Aram. ¨bdt, Ìzyt (with Altheim and Stiehl, e.g., Diearamäische Sprache, 298), but that the -T was in the course of time reinter-preted as a phonetic complement (-t) and so could be added to arbitraryforms. In the same way, I assumed that Middle Persian HZY-T-N was Ìzytwith the addition of the “enclitic” subject 1st sing. personal pronoun in re-duced form -en: < Ìzyt-nˆ.28 As the reduced form -en of the 1st sing. personalpronoun in Syriac is only attested with participles, we can assume that its usewith finite verbs was an Iranian practice.

The main types of verbal heterograms were thus:29

Aramaic Parthian Middle Persian

Type I active participle: AHD-W “take” OHD-WN“take” (OBYD-WN “do, make”)

HZYE “see” BOYH-WN ”seekMQBL-W “receive” MKBL-WN “receive”

Type II imperfect: YNTN-W “give” YKTL-WN “kill”(YKTYB-WN “write”)

Type III perfect: AHD-T/t “taken” OHD-WNHZY-T/t “seen” HZY-T-N30

HQAYM-t “placed” HNHT-WN “place”Type IV passive participle: KTYB-t “written” (YLYD-WN “be born",

etc.?)

298 ARAMAIC IN IRAN

28 This is not, as far as I can see, explicitly stated by Altheim and Stiehl, but follows logicallyfrom their other suggestions, that BNY-T is 1st sing. b¢ne† and that mˆytyn contains the re-duced form of the 1st sing. pronoun.

29 These various types were described in some detail by Herzfeld (Paikuli, 52-59) andHenning (“Mitteliranisch”, 35-36). As far as I can tell, Henning ignored the work ofHerzfeld, whose book on Zarathustra he had just demolished in his Zoroaster. Politician orWitchdoctor? (London, 1951).

30 In Book Pahlavi (from ca. 6th century C.E.), in which the shape of the letters w and n(Psalter and ) had merged into , this type was eventually read as HZYTWN-.

Page 17: Aramaic in Iran

Several forms still need individual explanations of details, of course.Thus, the Parthian verbs with weak third radical had the form HZYE for thepresent, which may be a conflation of Ìzy/Ìzh = Ìaze, from Ìaze-h(u)(?), orsimply HZY with the -E of Ìzh, etc.

The Middle Persian verbal heterograms, being reduced to one form forboth present and past stems, are a priori more likely to incorporate analogicalforms and conflations of original differentiated forms. The fact that almost allMiddle Persian verbal heterograms end in -N indicates that the -N in some ofthem is analogical.31 Thus, the final -N in the type OBYD-W-N, MKBL-W-N is probably analogical from the 3rd plural forms of the type YKTLWN < yq†lwn. One possible scenario is the following. (1) Forms of the type*OHD-W “take” < ˆÌdw present participle + pronoun or 3 plur. perfectacquired the -N of the imperfect yˆÌdwn. (2) With two types ending in -WN(OHDWN and YKTLWN) this ending would have started spreading to otherforms, such as *BOYE “seek” < ba¨e (or. sim.), which became BOYHWN“seek". So many prototypes are possible that I refrain from reconstructing animaginary proto-system.

Parthian forms of the type HZY-T, if analyzed as HZY-t, can be eitherType I or IV. For simplicity’s sake I prefer to interpret them the same way asOBD-T (or OBD-t), which cannot be Type IV (*OBYD). Middle Persianforms of the type YLYD-WN “be born” can be either Type I or IV. The formYKTYB-WN can be a conflation of *YKTB-WN and *KTYB, but also befrom yiÈteßun written plene like ˆthÌsynn = e†haÌsen- in the Asoka inscription(see above).

It is, finally, also possible that the -N in HZYTN is the result of the sameanalogy, rather than an original Aramaic 1st sing. pronoun.32

On this basis we can set up the following basic system of correspondencesbetween the verbal heterograms and the Iranian forms they represent (I giveonly one example of forms other than the 3rd sing.):

PRODS OKTOR SKJÆRVO 299

31 Indeed, the attested Middle Persian verbal heterograms without -N stand out like sorethumbs in the system, and some of these acquire -N before our eyes, e.g., OSTE “eat” (SPsI) > OSTE-N (Paikuli), YCBE “wish” (Paikuli) > Book Pahlavi YCBE-N.

32 My recent discovery that the 1st singular imperfect is attested in Middle Persianinscriptions with the ending -en (“L’inscription d’Abnun”), opens the possibility of a reinter-pretation of the phonetic complement as part of the heterogram: HZYT-n > HZYT-N, as inParthian OBD-T > OBD-t, etc.

Page 18: Aramaic in Iran

Present (and imperfect) Simple past

PARTHIAN

Intransitive: AZLW = sawed “he goes/is going” AZL-t (AZLT) = sud“(he is) gone”

AZLW-m = sawam “I go/am going” AZL-t (AZLT) HWY-m = sudhem “I am gone”

ATYE = ased “he comes” ATY-t (ATYT) = agad“(he is) come”

Transitive: OBDW = kared “he does/is doing” W-s OBD-t (OBDT) =u-s kerd (lit.)“by him done” = “he did”33

OBDW-m = karam “I do/am doing” W-m OBD-t (OBDT) = u-mkerd (lit.) “by me done” =“I did”

HZYE = wened “he sees” W-s HZY-t = u-s did “he saw”HQAYMW = awested “he places” W-s HQAYM-t (HQAYMW-t)

= u-s awestad “he placed”YNTNW = dahed “he gives” W-s YNTN-t = u-s dad

“he gave”KTYB-t = nibist “written”

MIDDLE PERSIAN

Intransitive: OZLWN = sawed “he goes/is going” OZLWN = sud“(he is) gone”

OZLWNm = sawem “I go/am going” OZLWN HWEm = sud hem“I am gone”

YATWN = ayed “he comes” YATWN = amad“(he is) come”

Transitive: OBYDWN = kuned “he does/is doing” AP-s obydwn = u-s kerd“he did”

OBYDWN-m = kunem “I do/am doing” AP-m klty = u-m kerd“I did”

HZYTN = wened “he sees” AP-s HZYTN = u-s did“he saw”

YKTLWN = ozaned “he kills” AP-s YKTLWN= u-s ozad“he killed”

300 ARAMAIC IN IRAN

33 Note that in the Parthian inscriptions the agent is less often expressed than in the Mid-dle Persian ones, cf. Sapur Hajjiabad above.

Page 19: Aramaic in Iran

In the inscriptions the phonetic complement -t is optional in the 3rd sing.present (indicative, subjunctive), while the 3rd sing. simple past hardly everhas a phonetic complement. In the Psalter and in Book Pahlavi the 3rd sing.present takes -yt and the 3rd sing. simple past -t:

Mid. Pers. inscriptions Psalter, Book Pahlavi3rd sing. present sawed OZLWN or OZLWN-t OZLWN-yt3rd sing. past sud OZLWN OZLWN-t

Note that all the simple past forms can also be used in passive (agent-less)constructions, in which case they must be rendered as passives, e.g., ParthianME OBD-t = ce kerd “which (was) done", MNW QQL-t HWYN = ke ozadhend “who were killed".

In view of the requirement that one Aramaic form must do duty for allthree persons in both numbers, we see that the Parthian heterograms as inter-preted above are a nearly “perfect” fit for the Iranian forms and that many ofthe Middle Persian heterograms can be interpreted as containing appropriateAramaic forms. What is perhaps most surprising is that the Aramaic hetero-grams for the simple past normally render the meaning of the Iranian con-struction, which is active, not the form of the construction, which is passive(see Skjærvø, “Remarks”, 221-223).

With the above theory, by considering the correspondences between bothform and function in both the Aramaic and the Iranian systems, I was able toexplain in a relatively simple manner both the forms the verbal heterogramstook in Iranian and their function there:

(1) Function: The system is based only upon the correspondence betweenthe functions of the verbal categories of the Aramaic verbal system: the im-perfect, perfect, and participle, and the functions of the Iranian verbal cate-gories: present-imperfect and simple past. Aramaic imperfects and active(present) participles represent the Iranian present-imperfect, while perfectsand (rarely?) passive (perfect) participles represent the Iranian simple past.

(2) Form: The heterograms can to a great extent be explained as actual“frozen” Aramaic forms corresponding to their functions.

(3) The development of such a system can be fitted into a plausible theoryof who made it, why did they make it, and how did they make it? (see theConclusion).

C. TOLL'S THEORY

A different explanation of the origin of the Aramaic heterograms in Iranianwas outlined recently by C. Toll, a student of Nyberg’s, in a presentation at

PRODS OKTOR SKJÆRVO 301

Page 20: Aramaic in Iran

the XXIV. Deutscher Orientalistentag, September 1988 in Cologne, published1990 (Toll, “Die aramäischen Ideogramme”).

In a letter to me he says that he wanted to replace the old method, which con-sisted in taking various forms “imperfect or perfect, 1st or 3rd sing. or 3rd plur.,active or passive participles in inexplicable mixture and helping out with anal-ogy", with “an explanation that is less arbitrary and a more systematic theory".This is alluded to in his article (p. 28), where he criticizes the “frozen form”explanations of the verbal heterograms by questioning the rationale34 behind thechoice of a 3rd plur. imperfect, such as YKTLWN- “kill",35 a 2nd plur. perfector participle + pronoun, such as HZYTWN-36 “see", or hybrids of imperfectsand participles with imperfect pre- and suffixes, such as YKTYBWN “write”and Y-KOYM-WN37 “stand", to express all the forms of the Middle Persianverbs. Similarly, he argues (p. 33) that, because a verbal heterogram (in theMiddle Persian inscriptions) can express both the past participle and the imper-ative [also, but unknown to Toll, the 3rd sing. present], we must conclude thatthe heterogram is not to be “understood” as perfect, imperfect, or participle.

We see here how the question of the origin and function of the heterogramsin the Iranian text is confused with that of their function in the Aramaic text.38

302 ARAMAIC IN IRAN

34 “Es ist schon schwer zu verstehen… noch schwieriger zu verstehen… Und noch schwie-riger … Ein solches „System“ scheint recht unbegreiflich–ja, man fragt sich, ob man hierüberhaupt von einem System sprechen kann” (p. 28).

35 On p. 34, he dismisses the possibility of reading the form as Aramaic meaning “onekills, they kill” on the grounds that “there is no reason to do so, because [according tohis theory] -KTL- is the perfect or root and Y- and -WN determinators” (“dazu gibt es keinenAnlaß, sondern… -KTL- ist die Perfektform oder die Wurzel als Ideogramm, und Y- und-WN sind die Verbaldeterminative”). On p. 32 he asks what the function of a YKTLWN“man tötet” would have in the Persian text, to him, no doubt, a rhetorical question implying,I suppose, that it would have none. In fact, the 3rd plural is regularly used in Persian andother Iranian languages to express agent-less statements like “one kills”.

36 Actually, this form is a late analogical deformation of HZYTN, which is unlikely to bea 2nd plur. form. We should also note that in Book Pahlavi, w and n merged into one sign andthat in many old manuscripts one has the impression that this type of verbs is to be read actuallyas HZYTN-, rather than as HZYTWN-, which may therefore be entirely non-existent. Astatistical investigation of the forms in the best manuscripts is necessary to settle this question.The pattern KKY-T-WN (p. 42) is therefore just a late variant of KKY-T-N.

37 The reason for transliterating this word with ¨ayn, rather than, e.g., with W(*YKWYMWN), is that the Parthian form is HQAYM-, which has A instead, and a corre-spondence Parthian A ~ Middle Persian O is seen also elsewhere, notably in Parthian ALHA“god, lord” ~ Middle Persian ORHYA (< ˆlhyˆ).

38 Note also on p. 33: “Der Umstand, daß Ideogramme ohne persisches Komplementsowohl für persisches Partizip Prät. und persischen Imperativ (vom Präsensstamm) stehen

Page 21: Aramaic in Iran

Obviously, the Middle Persian heterograms are no longer, in the Middle Persiantext, to be understood as Aramaic 3rd plur. imperfect forms or participleswith attached 2nd plur. pronouns, but this does not mean that they cannot bederived from such forms.39

Toll wants to explain the form taken by the Aramaic heterograms in MiddlePersian40 by a proposed “theoretical” derivation, assuming that they originatedas “ur-ideograms” but were provided with various suffixes and prefixes in-dicating more explicitly their grammatical function. He goes on to state histheory, which is “functional". This term as used by Toll does not refer to thesyntactic function of the heterograms (if I understand correctly), but the lexicalcategories of “noun” and “verb” and the morphological categories of “strong”and “weak” verbs. According to him the forms were assigned to the hetero-grams by the scribes specifically to distinguish clearly between these“functions",41 and he is investigating how the heterograms express these“functions” and trying to explain them on this basis.42 Thus, also in hisconclusion, he states that his theory is based upon the “functions” of theheterograms in the Persian text43 and that the heterograms represent a logical,although not quite consistent, system designed to distinguish between categories

PRODS OKTOR SKJÆRVO 303

können, zeigt… daß das Ideogramm, die aramäische Verbalform [my italics], tempusneutralist, weshalb es kein Grund gibt, die Ideogramme als Perfect, Imperfect oder Partizip zu ver-stehen”.

39 Note also, on the same page, “… man kann zur Not die meisten Ideogramme aus Formendes Mandäischen oder Neusyrischen erklären, aber das Ergebnis sind Formen, die für denGebrauch der Ideogramme völlig belanglos sind [my italics], und zuletzt bleiben immer nochFormen… die auf keinen Fall mit irgendeiner bekannten aramäischen Verbalform identifiziertwerden können… [oder] die ebenfalls nicht aramäisch gelesen werden können [my italics]”.

40 “Der Zweck des Aufsatzes ist, die Formen zu erklären, welche die aramäischenIdeogramme im Mittelpersischen haben” (p. 28).

41 “Demgegenüber möchte ich behaupten, daß die Ideogramme überhaupt nich aufAramäisch gelesen werden sollen oder können, weil sie keine aramäischen Formen darstellen,weder reichsaramäische noch ostaramäische, sondern sämtlich von den Schreibern fabriziertworden sind, und zwar nicht aus mangelnder Kenntnis des Aramäischen, sondern zu ebendem Zweck, zu dem man überhaupt Ideogramme benutzte, nämlich der größeren Deutlichkeitund Verständlichkeit halber. Meine Theorie geht von der Funktion der verschiedenen Formender Ideogramme aus. Diese Funktion is nicht, Tempora und Personen anzugeben… [sondern]vielmehr, Nomina und Verben, Stammformen und schwache Verben zu identifizieren” (p. 28,bottom; my italics).

42 “Ich will nun untersuchen, wie die Ideogramme die von mir vorausgesetzte Funktionwahrnehmen, und die Formen von dieser Funktion her erklären” (pp. 28-29).

43 “… eine Theorie… die von der Funktion der Ideogramme im persischen Text ausgeht”(p. 40).

Page 22: Aramaic in Iran

of words and between verbs from different stems and with different meaningsin order to ease the reading and understanding of the text.44

My criticism of Toll’s theory addresses the following three issues: (1) thechoice of material, (2) the omission of references to several publications onthe subject, and (3) his theory itself.

Toll’s decision to study the Middle Persian heterograms on the basis ofNyberg’s Manual II and the Frahang i pahlawig,45 which encode late BookPahlavi usage (ca. 7th-9th centuries C.E.), instead of choosing as his basisthe oldest texts, that is, the inscriptions and the Pahlavi Psalter (3rd- ca. 6thcenturies C.E.), is surprising. One of the disadvantages of using the late materialis that the Book Pahlavi script is extremely ambiguous, as opposed to theearly Parthian and Middle Persian lapidary and Psalter scripts, which containvery few ambiguities. The systems of phonetic complements in these variouscorpora also differ to a great extent (see Skjærvø, “Verbal Ideograms”). TheBook Pahlavi evidence therefore gives a rather different picture from that ofthe inscriptions and the Psalter.46

Toll ignores some crucial secondary literature, including Herzfeld, Paikuli(1924), Gignoux, “Étude des variantes textuelles” (1973), Brunner A Syntax(1977), Humbach-Skjærvø, Paikuli (1983), and Skjærvø, “Case” (1983) and“Verbs in Parthian” (1986), which leaves him unaware of important pointsabout the syntactical functions of the heterograms, especially their correlationwith the Iranian tenses (see above). His question, “how often and by whatprinciples ideographic or phonetic spellings were chosen” (p. 27)47 had in factalready been answered in part by Herzfeld and myself.

304 ARAMAIC IN IRAN

44 “… ein logisches, wenn auch nicht konsequent durchgeführtes System dar, um die ver-schiedenen Wortarten zu kennzeichnen, um Verben verschiedener Stammformen und damitverschiedener Bedeutung zu unterscheiden und um die Lesung und das Verständnis einesTextes auch sonst zu erleichtern” (p. 40).

45 “Ich gehe dabei von dem System aus, wie es im Buchpahlavi vorliegt, mit NYBERGS

Manual. 2. Glossary und Frahang als Quellen” (p. 28). 46 Thus several statements on p. 37 are based on insufficient material: It is not true that in the

Middle Persian inscriptions the heterogram without phonetic complement is mostly used for “theparticiple in the preterite”, nor is it true that there is only one heterogram without phonetic com-plement used for the imperative. The statement that only in Book Pahlavi does the heterogramwithout phonetic complement used for the imperative become more frequent is meaningless, asthe Book Pahlavi texts cover a much wider range of prose than the inscriptions.

47 “Es gibt in diesem Zusammenhang noch ein Problem, das ich nicht näher beachtet ge-sehen habe: Wörter, die mit Ideogrammen geschrieben werden, können daneben auch phonetischgeschrieben werden. Es dürfte interessant sein zu wissen, wie oft und nach welchen Gründenideographische oder phonetische Schreibung gewählt wurde”.

Page 23: Aramaic in Iran

Toll elaborates his theory by discussing consecutively nominal, verbal, andpronominal heterograms. The discussion of the verbal heterograms is sub-divided into that of strong and weak verbs. His conclusions about the verbalheterograms are set out in table form at the end of the article. My critiquefollows this outline, except that I give my remarks on the table in my discussionof the verbal heterograms.

Nominal heterograms

Toll first discusses the nominal heterograms, many of which have theending -A. This he interprets as invented by the scribes to distinguishnouns from the verbs: because mlk could mean both “king” and “rule",MLKA was chosen as the heterogram to designate “king". The nominalheterograms in -E he interprets as an alternate spelling for the determinatestate, that is YDE “hand” for Aramaic ydh = ydˆ.48 The ideograms forfamily terms: ABY “father", etc., he interprets, traditionally, as originalvocatives containing the Aramaic possessive pronoun -i which becamenormal forms.49

This theory does not explain why nouns for body parts tend to take theending -E rather than -A in Middle Persian (as opposed to Parthian!), whythe nouns for family terms end in -Y, and why some nominal heterogramshave no ending at all. It also does not explain why the Iranian languages thatused the heterographic system of writing differed in their use of forms withno ending, -A, or -E.

We see that to explain the forms in -Y and those without ending, Toll hasrecourse to the “frozen forms” theory, which he is criticizing, and explainsthem by their function in Aramaic. But why should the inventors have usedvocative forms to express forms that in the administrative documents mostoften functioned as anything but vocative or have been concerned about astatistical preponderance of construct forms50 in a small set of nouns?51

PRODS OKTOR SKJÆRVO 305

48 Support for the interpretation of the final -H as a spelling variant of -A can be found inParthian scribal practice. In fact, Mid. Pers. YDE = Parth. YDA, Mid. Pers. LOYSE “head”= Parth. RYSA, a fact of which Toll seems to have been unaware.

49 “… alle diese Wörter können als ursprüngliche „Vokative“ betrachtet werden, die zurNormalform des Wortes geworden sind” (p. 29, bottom).

50 “… einige Nomina, die im Aramäischen meist im St. cstr. stehen” (p. 31). This ex-planation also breaks down: “Warum die folgenden Wörter kein Determinative haben, istaber schwer zu erklären…” (p. 31).

51 On pp. 30-31, Toll discusses the heterogram OBDk/OBDK = bandag, which Toll, pos-sibly correctly, derives from Aram ¨aßdaÈ “your servant”. The only attestations in the inscrip-

Page 24: Aramaic in Iran

Finally, the theory does not explain forms such as Parthian MLKYNMLKA “king of kings", with an Aramaic plural form mlkyn.

Why there was a need to distinguish nominal from verbal heterograms isnot made clear. As a matter of fact, very rarely (if at all) do we find Aramaicroots used for both noun and verb, as vaguely implied by Toll. Thus, there isno verbal heterogram from mlk, either in Parthian or Middle Persian, and no“couple” MLKA ~ *MLKWN, *YMLKWN, both differentiated from anundifferentiated ur-ideogram *MLK meaning both “king” and “rule". Thenumber of Iranian nouns (written with heterograms) derived from verbs isprobably close to nil (or non-existent). Normal verbal nouns if written withheterograms rather than phonetically are indicated by phonetic complements,for instance, HZYTN-sn = wenisn “seeing, sight", HZYTN-tˆl = didar52

“visible”; there is no “couple” *HZYA, *HZYE ~ HZYTN. Finally, even if there were a substantial number of such “couples",

the normal syntax of the Iranian sentence (SUBJECT/AGENT… VERB) woulddistinguish more than clearly enough between nouns and verbs. This fact,too, would seem to obviate the scribes’ need for inventing a way ofdistinguishing between nouns and verbs. On the other hand, adjectives, whichfrequently do occupy the same position in the sentence as verbs, are notdistinguished in any way, and the adjectives of the form KKYK (SPYL, KBYR,etc.) thus coincide completely with Toll’s verbal pattern KKYK (p. 41).

Verbal heterograms

To explain the verbal heterograms, Toll basically assumes that it would havebeen natural for the scribes to chose the 3rd sing. perfect, the “fundamental”form of the Semitic verb, the one which in the script coincided with the root,as verbal heterogram;53 these verbal ur-ideograms then received determinatorsthat characterized them as verbs, and, since two of the characteristic featuresof Aramaic verbs were the 3rd plur. endings -u and -un and the 3rd pers.imperfect prefix y-, these were consequently the determinators chosen to

306 ARAMAIC IN IRAN

tions actually do support this meaning for Middle Persian, as well. For the passage (SPs II 7-11)see, e.g., Skjærvø, “Verbal Ideograms”, 343). As for his query p. 30 whether bandag is attestedin such a function, actually, OBDK/k is attested in this meaning in SPs II line 16. The “hetero-gram” AZk quoted in this context is probably not a heterogram, however, but an obsolete Iranianword: Avestan aza-, see Skjærvø, review of Nyberg, Frahang i pahlavik, 97.

52 The word seems to have been analyzed, logically, as did + ending -ar (HZYTN-t-ˆl). 53 “Die Grundform des semitischen Verbs, diejenige, welche in der Schrift mit der

Wurzel zusammenfällt, is bekanntlich die 3. P. M. Sg. Pf. Es wäre natürlich gewesen, wenndie Schreiber diese Form für das Verbalideogramm gewählt hätten” (p. 32).

Page 25: Aramaic in Iran

characterize the forms as verbal ideograms. This does not mean, according toToll, that the ideograms actually represent the corresponding Aramaic forms:the similarity is a coincidence, and the ideogram expresses only the idea ofthe action.54 This is true at the time of our inscriptions and later, but was italso true when the forms first were chosen or became fixed?

The notion of an ur-ideogram developing into the attested forms by ac-cretions is not new; it is nothing but Henning’s “onion” simile in a moreelaborate form: “one must peel a form like HZYTWN like an onion in orderto recognize the Aramaic ur-ideogram HZY".55

On the basis of these assumptions, Toll suggests the following stages ofdevelopment:

1. Ur-ideogram: KKK, only in Parthian.56

2. Addition of the element -W to distinguish them from nouns [my italics],in Parthian.57

3. Expansion of KKK-W to KKK-WN58 for greater clarity [my italics]. 4. Addition of Y- to the form in -WN, for maximum clarity [my italics].

In the case of Pa¨¨el forms, the 3rd sing. masc. of which is not distinguishedfrom the Qal, the prefix M- from the participle was chosen rather than Y- todistinguish the two forms (p. 38).

PRODS OKTOR SKJÆRVO 307

54 “Wenn es auch die Form yiq†elun auf Aramäisch gibt, so ist das nur ein Zufall, und alsIdeogramm drückt das Wort nur die Idee der Handlung aus” (p. 32). From this point of view theaspect of the action expressed by the heterogram is of course irrelevant, as stressed reapeatedlythroughout the article. Nevertheless, occasionally Toll invokes “punctual” versus “durative”aspect to explain the forms, e.g., p. 35 (bottom), 36 (middle). It may be pointed out thatParthian and early Middle Persian do not distinguish between these “aspects” of the action,see Skjærvø, “Remarks”, 222-225.

55 “… eine Schreibung wie ÎZYTWN muss man zwiebelgleich auswickeln, um das wirk-lich aramäische Urideogramm ÎZY erkennen zu können” (Henning, “Mitteliranisch”, 36).

56 That is, if we accept that the forms of the pattern KKK-T are originally KKK-t withIranian phonetic complement, rather than Aramaic 1st sing. perf., later reinterpreted as con-taining a phonetic complement (see above). An ur-ideogram KKK alone is thus not attestedin Parthian (or Middle Persian).

57 Toll adds, perhaps also in the Middle Persian inscription, referring to his suggestion thatthe final -N in Middle Persian verbal ideograms is a misreading for -W (see below). As a matterof fact, there were Middle Persian verbal ideograms ending in -W: the one attested is YHY-TYW, which Toll does not mention. – The corresponding Parthian Hap¨el pattern H-WKK isrepresented by hwsr-t in some Arameo-Parthian documents, while, H-WKK-W is Gignoux’s“[Î]WD¨N[t]”, for which we must read [H]WDOY[W].

58 There are no Parthian forms of this pattern, as “AZLWN” in Gignoux, Glossaire, is amistake for AZLt.

Page 26: Aramaic in Iran

The various forms representing the ur-ideograms of the strong and weakverbs and their subsequent modifications are set out at the end of the articlein a table of patterns of heterograms in the Parthian and Middle Persianinscriptions and the Frahang i pahlawig, together with the number of hetero-grams attested for each pattern.59 The inscriptional forms are cited fromGignoux’s Glossaire, which leads to numerous mistakes, especially sinceHumbach-Skjærvø, Paikuli, was not consulted. Evaluating this table is difficult,since the forms themselves are not quoted and so have to be hunted down inGignoux’s Glossaire. Many of the patterns are limited to single and/or doubtfuloccurrences or to late Book Pahlavi forms more developed than the corre-sponding forms in the inscriptions or the Psalter (see the notes above andbelow). The table therefore presents many pitfalls for the unwary Aramaist.

Following are comments on some details of general interest.Toll’s suggestion that the final -N without preceding -W- (-WN) in Middle

Persian ideograms is actually -W (pp. 34-35), implies that this -W has beenmisread as -N by a century of Iranian scholars. It is obviously free fantasy, aseven a cursory glance at any inscription (Middle Persian = -WN) or apage of the Psalter will show (Ps. = -WN).60

The suggestion that the ending -WN is an expansion of -W must be basedupon the assumption (not expressed by Toll) that the Middle Persian systemrepresents a development of the Parthian system. The Middle Persian systemas a whole, however, cannot be derived from the Parthian one, and there istherefore no reason to assume–on the basis of the attested forms–that -WN isfrom -W with the addition of a clarifying -N.61

308 ARAMAIC IN IRAN

59 The Psalter forms are not included. 60 Similarly, the suggestion that Book Pahlavi heterograms with only -N following -E-

or another -N- show a reduction of -WN to -N while leaving the heterogram ending in twovertical strokes (BP - = -EN = -MNN and - = -NN) ignores the fact that the heterogramsin question in the Middle Persian inscriptions originally ended in -E only and -NN, respec-tively: OSTE “eat” (later OSTEN) and YHSNN “hold, have” (not *YHSNWN). The latterform is the only example of the pattern Y-KKK-N (p. 41); neither the old theory nor Toll’snew one can explain this form well, which does not agree with any known Aramaic form(speculatively: perhaps conflation of *YHSN < *y(¢h)aÌsen “he holds” and *MHSNN <*m(¢h)aÌsen-en “I am holding”). KKK-n refers to THNN- “grind”, only(?) attested in Fra-hang i pahlawig, chap. 19, which tells us nothing about what its original (intended) form mayhave been.

61 Toll’s suggestion that the ending -WN ( -) in Book Pahlavi was no longer understoodas such, only as two final strokes, is rendered unlikely by the fact that the traditional pro-nunciation of these forms as recorded in the 18th century by Anquetil-duperron was still with-un, see below.

Page 27: Aramaic in Iran

The addition of Y- to the form in -WN for maximum clarity is justifiedby reference to nouns and adjectives ending in -WN or beginning with Y-(pp. 34-35).62 Such nouns are so rare, however, that they can hardly haveexerted enough pressure on the, according to Toll, already differentiated verbalideograms for them to be further differentiated. (It would have been mucheasier to change the few nouns and adjectives in question.)

Toll’s explanation of the heterograms in M- as intentional differentiationof the Pa¨¨el forms from those of the Qal, is subject to the same criticism ashis explanation of the distinction between nominal and verbal heterograms.There are very few (if any) couplets that are distinguished only by the pre-fixes Y- ~ M-. Indeed, the attested couplets are distinguished differently, e.g.,ZBNWN (not *YZBNWN) “buy” ~ MZBNWN “sell".63

To explain the forms that do not conform to these patterns, for instance,Parthian KKYK and YKKYKW,64 Middle Persian KKYKWN andYKKYKWN, Toll constantly has recourse to the “frozen forms” theory hedisparages and intends to replace.65

Toll does not mention the forms of the copula, which are clearly derivedfrom individual Aramaic forms: Parthian HWY- < hawe for the present stem(HWY-m “I/we are”), HWE < hwa for the imperfect (Skjærvø, “On the MiddlePersian Imperfect”, forthcoming), AYTY for the existential verb “there is",

PRODS OKTOR SKJÆRVO 309

62 Nominal heterograms in -WN include PKDWN paymar “appointment, assignment”and HWBDWN wani/wany “destruction”, both used in verbal constructions with kardan “do,make” and budan “be, become”, meaning “appoint/be appointed” and “destroy/be destroyed”;they may be original verbal heterograms, cf. Parthian HWBDW-t “destroyed”. The form“magun” cited p. 35 1st line with a query is the oblique plural of “magu” “Magian”; I donot understand why it is mentioned here. – Heterograms in Y- quoted by Toll (p. 35 top) areYDE “hand”, YMA “sea”, YRHA “month”, and YWM “day”!

63 My own explanation of this couple is as derived from present participles zaßen-(h)uand m¢zabben-(h)u, see above.

64 KKYK is found in Parthian KTYB- “written” (only with phonetic complements:KTYB-t, KTYB-tn). Parthian Y-KKYK-W refers to Gignoux’s “Y¨RYBW-” and “YDRYKW-”,which are the same verb, but whose reading and meaning are quite uncertain, see Humbach-Skjærvø, Paikuli, Part 3.2, 65-66. The bracketed pattern KKYK-A (p. 41) is Armazi ¨bydˆ,which Altheim (Die aramäische Sprache, 42) proposed to read as ¨byd(w)ˆ, with -w- insertedabove the line.

65 He invokes “interference” (that is, analogy) also in other instances, as in the case ofthe one verb of the pattern KKK-L-WN, which is tentatively explained as “vielleicht Inter-ferenz” of the many [Aramaic!] verbs that have L as third radical. The verb in question isPahl. HCDLWN- = drun- “to reap (corn)”, where the interference is clearly from the Pahlaviverb itself (see Skjærvø, review of Nyberg, Frahang i pahlavik, 98).

Page 28: Aramaic in Iran

HWYN for the 3rd plur. present; Middle Persian HWE < hawe for the presentstem (HWE-m, etc.), HWYTN for the imperfect stem, AYTY for the exis-tential verb, negated LOYTY.

Strong and weak verbs

According to Toll, the inventors of the heterograms decided to make adistinction between strong and weak verbs, using a different selection principleof which basic forms to adopt in the two cases. Thus, for the strong verbsthey “chose", as we have seen, the 3rd sing. perfect = the root. Coming to theweak verbs, however, they(?) noticed a problem,66 namely that the 3rd sing.perfect ≠ the root: they no longer had three radical consonants, only two, theweak radical having been replaced by a long vowel. Why this should havebeen a “problem” to the inventors, we are not told. The problem is rather thatof Toll, since he has postulated that the ur-ideogram is the 3rd sing. perfect,not the root. The inventors of the heterograms could have just taken the 3rdsing. perfect and modified it the way they did the strong verb, making formsof the type *KKW, *KKWN or *YKKW, *YKKWN, which would have beeneasy to recognize and read, for instance, Parthian *QMW/Middle Persian*KMWN and *YQMW/*YKMWN or *HZW/*HZWN and *YHZW/*YHZWN.It is true that the ur-ideogram of verbs like Ìzy would end in -A or -E (*HZA,*HZE), which (by Toll’s theory) would make them indistinguishable fromnouns in -A or -E,67 but surely, if there was such an ur-ideogram, it wouldhave been easier and more consistent just to add Y- and -W/-WN, giving*(Y)HZEW, *(Y)HZEWN (or *(Y)HZHW, *(Y)HZHWN), forms that wereperfectly recognizable and unambiguous, rather than replacing them withforms in -YT: HZYT, HZYT-N.

Since the actual forms are not those required by Toll’s basic assumptions,his theory again breaks down, and, once more, he has to have recourse toother, sometimes quite complicated, explanations, the same as the “frozenform” theory he wants to replace.

The complicated explanations he has to provide are in themselves proofthat his theory is not up to the task. Note, for instance, that the form YATWN“come” according to Toll has nothing to do with Aramaic yˆtwn, but is from an

310 ARAMAIC IN IRAN

66 “Wenn das Ideogram ein schwaches Verb wiedergeben soll, ergibt sich das Problem,daß das Pf. der med. und tert. w/y nur zwei der drei Radikale wiedergibt” (p. 36, top).

67 They did not mind Middle Persian HWE, copula, however.

Page 29: Aramaic in Iran

ur-ideogram -AT-, with “curious” loss of the third radical (present in BookPahlavi Y-HYTY-WN “bring”) and the standard additions Y- and -WN.68

As an example of a form needing serious work to explain we may take Y-KOYM-WN “stand",69 which Toll first derives from the present participleqaˆim or qayim, with the addition of the “ideogram-markers". The reasonwhy this could happen is according to Toll that the meaning of Aramaic qamis punctual: “stand up", while Persian estadan is durative: “stand", andcould be denoted by the Aramaic present participle.70 As his basic assumption,however, is that the scribes made a conscious choice to use the 3rd sing. perfect,such considerations should not enter their discussion.71 As an alternate ex-planation he suggests that -KAYM- is for *-KYM- and that, as *YKYMWNwould be ambiguous in Aramaic (Qal or Pa¨¨el), an -A- was added before the-Y- in order to show that the heterogram stood for the Qal qam rather thanthe Pa¨¨el qayyim, although, as the form was to be used in Iranian, not Aramaic,such concerns would be irrelevant.72 As a matter of fact, in Middle Persian

PRODS OKTOR SKJÆRVO 311

68 The Parthian form ATYE, pattern KKY-E, is characterized as “mixed form” (“Misch-form”, p. 43, see also p. 37), which does not explain what was in the minds of the inventors.Other examples of weak verbs with prefix Y- include Parthian forms of the pattern Y-KK andY-KK-E (p. 42) represented by YBO-t, YBO-E “seek” (rhyming with YDO-t, YDO-E“know”) and YHW-t, YHW-E “be(come)”. The Parthian pattern Y-KKA-, Y-KKA-E is re-presented only by YMQA-t, YMQA-E “arrive”. The Middle Persian pattern Y-KK-E(-N/-WN) (p. 42) is represented by YCB-E “wish”, for which Book Pahlavi has forms that can beread as YCB-E-N or YCB-E-WN. Forms from the Hap¨el of weak verbs are the patterns H-KK, H-KKY, H-KKY-W, represented by Parthian HYT-t = HYTY-t “bring” and Parthianand Middle Persian HYTY-W. The Middle Persian pattern H-KK-WN is represented in theFrahang i pahlawig, chap. 21, by HCGWN, which (if correct), is probably a graphic variant ofYHYTYWN-. The pattern Middle Persian Y-H-KK-WN is represented only by YHMTWN“arrive” (intransitive!) = Parthian YMQA-. The Middle Persian patterns H-KKY-WN andY-H-KKY-WN are represented by Book Pahlavi HYTYWN “bring” for inscr. HYTYW andBook Pahlavi YHYTYWN.

69 Not *YKAYMWN, as it is consistently cited by Toll. YKOYMWN is the traditional tran-scription of this form, based on the relationship with Parthian HQAYM- and assuming ParthianA ~ Middle Persian O (as in Parth. ALHA ~ Mid. Pers. ORHYA, Mid. Pers. AYTY ~ LOYTY,etc.), but the form can of course also be read as YQWYMWN, for instance, which could be aconflation of the Imperial Aramaic imperfect yqwmwn and the present participle qym.

70 As a matter of fact, estadan means both “be standing” (Germ. “stehen”), durative, and“ go and stand somewhere” (Germ. “sich (hin)stellen”), punctual.

71 Toll also invokes aspect in his explanation of heterograms of the pattern -KKYK- toexplain them as being from the perfect passive participle, another break with his theory.

72 Other forms of the Parthian patterns KAYK and KAYK-W (p. 41) include QAYL-t andQAYL-W, QAYM-W (mistake for HQAYM-W?), and SAYL-W, all in the Nisa documents; thebracketed pattern KAYK-YN is Armazi mˆytyn, on which see the text of the inscription,

Page 30: Aramaic in Iran

there was no problem, as the normal verb for “place” is HNHTWN- = nih-ad,while the (rare) causative esten- “make stand > place” of est-ad is expressedby adding the causative suffix to the heterogram: YKOYMWN-yn-.73

Pronouns

Toll ends his exposition of the Aramaic heterograms in Iranian with abrief discussion of the pronominal heterograms, LY “I, me", and LK “you".Instead of interpreting these forms as the actual Aramaic forms, Toll againprefers to see the initial L- as a determiner that characterizes these words aspronouns. The case of the pronouns is different from that of the nouns, how-ever, since BYTA without -A is still a word, while LY and LK without L- arenot words. His arguments are also unclear. He first says that the reason for thischoice could be that the corresponding Persian pronouns man “me” isoblique case, and to oblique (as well as direct) case. He then goes on to statethat this does not explain why the Aramaic oblique case forms were chosenand that the L- does not denote the oblique case, which would have beenpointless in the Persian text.74

Since the direct and oblique cases of the pronouns were distinguished inPersian (as implicitly admitted by Toll, too), however, why should not thescribes chose Aramaic forms accordingly? Toll’s theory does not account forthe direct form of the 1st sing. personal pronoun ANE, which he does notmention. Clearly, with a system ANE = an “I” (direct case) versus LY = man(oblique case), it is hard not to conclude that the forms were chosen becauseof their functions in Aramaic.75 Toll does mention that there are instances of

312 ARAMAIC IN IRAN

above. The Parthian patterns H-KAYK and H-KAYK-W are represented by HQAYM- andHQAYM-W. Note that in the Nisa documents Hap¨el forms are spelled with initial he, but inthe inscriptions with Ìe†; Toll appears to have counted individually hqˆym- (with he) andHQAYM- (with Ìe†). The Parthian pattern H-KYK (p. 43) is represented only by Gignoux’s“HRYMt”, Nisa, of uncertain meaning, and Y-T-KYK (p. 43) by Nisa YTKYN-t andYTKYN-W. We have to wait for the publication of more of the Nisa corpus to evaluate theAramaic elements in these documents, however.

73 On these verbs see also Humbach-Skjærvø, Paikuli, Part 3.2, 22-23. 74 “Das erklärt aber nicht, warum für die Pronomina… die Form im cas. obl. gewählt

wurde, d.h. warum die aramäischen Pronominalideogramme mit L- eingeleitet werden. DieErklärung ist wahrscheinlich ähnlich wie bei den Nominal- und Verbalideogrammen: L- be-zeichnet nich den cas. obl., was im persischen Text zwecklos wäre” (p. 39; my italics).

75 The 2nd sing. LK stands for both the direct and oblique cases, but that is because thetwo forms merged in Persian. In Parthian LK (oblique case) is still distinguished from ANT(direct case). For the use of case forms of pronouns in Parthian and Middle Persian seeSkjærvø, “Case”.

Page 31: Aramaic in Iran

pronouns without L-, for instance ZNE and LZNE; what he does not say isthat these two forms represent different Persian pronouns: ZNE = en, LZNE= im, both “this".

A much simpler interpretation of the attested forms would therefore bethat, as the heterographic system was materializing, the scribes took formsthat were available and assigned them to the Persian words closest in meaning.Note that Persian ed, also “this", was originally expressed apparently byeither HNA or LHNA, leading to the disappearance of LHNA as redundant.

We may finally note that in the case of the demonstrative pronouns, thedifferent Iranian languages went different ways, e.g., in Parthian ZNE = imand ZK = ed, both “this” (Mid. Pers. ZK = an “that”), LH-w (or LHW) = ho“he, that” (Sogdian ZK = xo “he, that”), etc., see Henning, “Mitteliranisch”,32-33.

Conclusion

As the reason for any alternative to an existing theory must be that thealternative theory provides a better description and explanation of the facts,we may ask whether Toll’s theory does so. The answer is clearly that it doesnot. It does not, for instance, explain the distribution of the Parthian forms in-T (or -t) and -W/-E between the past and present stems and the restriction ofMiddle Persian forms like OBYDWN to the present stem, facts the author isnot aware of, but which were described by Herzfeld in 1924.

The theory that the forms of the heterograms were chosen to distinguishbetween nouns and verbs and strong and weak verbs breaks down over thefact that there was no need for such a formal distinction.

The theory of the “onion-like” accretion process itself, in Toll’s opinionconsciously invented by the scribes, breaks down in that aberrant heterogramscannot be explained by it without recourse to a variety of ad hoc explanations(like the old theory).

It is not obvious (no sarcasm intended) how the “old” method of “takingperfects, imperfects, and participles and helping out with analogy” is lessadequate than the method of taking perfects (which are not perfects), addingthe grammatical elements of the imperfects (without thereby making theforms imperfects), and helping out with analogy from participles.

Finally, the basis of the theory, that the heterograms were consciouslyinvented by the scribes, brings up the question of how exactly the hetero-grams were “invented". We know that the scribes were an important part ofthe central and local administrations, as they are frequently mentioned in theoldest Sasanian inscriptions, and the homogeneous and consistent writing

PRODS OKTOR SKJÆRVO 313

Page 32: Aramaic in Iran

system we can observe in both the Parthian and Middle Persian texts mustclearly be the result of conscious policy-making. The step to assuming thatthey based their scribal policy on an abstract grammatical analysis of Aramaic,as assumed by Toll when he posits that the scribes chose the 3rd sing. perfectbecause it is the basic form of the Semitic verb, coinciding with the stem, isvery doubtful, however, as we do not seem to have any evidence from thistime (2nd-1st cents. B.C.E.) for the kind of grammatical science needed formaking such an analysis. Thinking in terms of grammatical abstracts comesnaturally to us, but to somebody without the kind of training we get, “grammar”consists of actual forms, not roots and stems. It is an experience any dialectresearcher will have had that when you ask for basic forms of the verbalparadigm, to us perhaps the infinitives, you will only get personal forms.Even the concepts of noun and verb may not have been familiar to them. Onthe contrary, if asked, they would just as likely say that mlk “he ruled” and“king” were the same kind of words, deducing from the basic meaning of thewords, a response language teachers are well familiar with from studentswithout grammatical schooling.

Imagining the scribal college of the Parthian administration, well versed inAramaic grammar, sitting together and deciding on the forms by taking theroot of the verb and then debating which prefixes or suffixes to add to it tomake it a heterogram implies a situation that clearly did not obtain at thetime. On the contrary, imagining the same college consisting of Parthians bynow fairly illiterate in Aramaic choosing from the forms most commonly usedin the actual documents at their disposal and the sample texts they must havelearned from their teachers, makes perfect sense, at least to me. The fact thatthe scribes no longer felt strongly that Ìzyt was a 1st sing. form and so usedit for all persons, brings to mind a famous Iranian archeologist (a European),whom I heard speak Persian with only one verbal form in his grammaticalrepertoire, namely the 1st plural (using it with all the personal pronouns),which was, because of its common use in a group of people, presumably theone he noticed most often.

Even if we admit, for the sake of argument, that an Aramaic scribal collegedid possess this kind of knowledge of its own language, how can we assumethat a Parthian or Persian college of scribes no longer sufficiently familiarwith Aramaic orthography to distinguish between h and Ì did so?

LEARNING HOW TO SPELL WITH HETEROGRAMS

How, then, did the scribes learn how to write with heterograms, we mayfinally ask? Clearly, once they ceased intentionally writing Aramaic and their

314 ARAMAIC IN IRAN

Page 33: Aramaic in Iran

knowledge of Aramaic faded, they could no longer call upon the actual Aramaicforms to help them. At a guess, I suggest they did what we did in school whenwe began learning foreign languages: they memorized the forms, probably inlists such as: “to write X, write Y". This method was extended to cover in-digenous words with “strange” spellings, as well. The earliest such list pre-served is the Frahang i pahlawig, which is in precisely this form: BYTA:xanag (“house”), etc. Later versions of the lists are found in early publicationsof Zoroastrian material. The following list of Pahlavi words in traditionalGujerati Parsi pronunciation is from Anquetil-Duperron’s Zend-Avesta (III,476-522). Anquetil-Duperron gives the Persian equivalents, which correspondto my transcription, and a French translation. In the far left column I havegiven the standard modern transliteration and in the second column the alter-nate reading of the Pahlavi graph which gave rise to the traditional “school”pronunciation. Note that pronouncing the Aramaic words as such in thelearning process does not, of course, make these words loanwords and part ofthe lexicon. It is just a learning devise.

modern “looks like” traditional Gujerati modern meaningtransliteration school pronunciation transcriptionˆwhrmzd ˆnhwmˆ anhuma Ohrmazd OhrmazdAYMT ˆdmt admat kay when?ÅNE76 ynh anâ en thisAB ab pid fatherABYtl abider pidar father (obl. case)HLKWNtn ˆlkwntn alkunatan baxtan to distribute,

giveHLKWNyt alkuned baxsed he distributes/

givesBRE bwmn boman pusar sonBRTE bntmn bonteman duxtar daughterBOYHWNstn bwyhwnstn bavihunastan xwastan to seekBKYWNstn bhwnstn bahunastan gristan to weepYNS∫WNtn ywsgntn josgonatan stadan to takeYCBENstn dcbmnstn dajbamunestan kamistan to wishMNYTWNtn mwytwntn mavitunatan osmurdan to rememberOBLWNtn nblwntn nabrunatan widardan to pass awayO∫YDWNtn wˆgwntn vagunatan kardan to do

PRODS OKTOR SKJÆRVO 315

76 The underscoring of single letters means that they are here written like another letter,e.g., B, K, or Z like Y and ZD, BY, and KY like A.

Page 34: Aramaic in Iran

REFERENCES

ALTHEIM, F. and R. Stiehl, Die aramäische Sprache unter den Achaimeniden, Vol. I,(Frankfurt am Main, 1963).

ANDREAS, F. C., and K. Barr, “Bruchstücke einer Pehlevi-Übersetzung der Psalmen”,SPAW, (1933), 91-152.

ANQUETIL-DUPERRON, A. H., Zend-Avesta, ouvrage de Zoroastre, Vols. I-III, (Paris,1771; repr. New York and London, 1984).

BOWMAN, R. A., Aramaic Ritual Texts from Persepolis, (The University of ChicagoOriental Institute Publications XCI, Chicago, 1970).

BRUNNER, C. J., A Syntax of Western Middle Iranian, (Delmar, NY, 1977).COWLEY, Aramaic Papyri of the Fifth Century B.C., (Oxford, 1923; repr. Osnabrück,

1967).DALMAN, G., Grammatik des jüdisch-palästinischen Aramäisch, (Leipzig, 1905; repr.

Darmstadt, 1960). GIGNOUX, Ph., “Étude des variantes textuelles des inscriptions de Kirdir. Genèse et

datation”, Le Muséon, 86, (1973), 193-216.GIGNOUX, Ph., Glossaire des inscription pehlevies et parthes, (Corpus Inscriptionum

Iranicarum, Supplementary Series, Vol. 1, London, 1978). HENNING, W. B., “Das Verbum des Mittelpersischen der Turfanfragmente”, Zeitschrift

für Indologie und Iranistik, 9, (1933), 158-253 (= Selected Papers, vol. 1, 65-160).HENNING, W. B., “Mitteliranisch”, in Handbuch der Orientalistik, I, IV, 1, (Leiden-

Cologne, 1958).HERZFELD, E., Paikuli. Monument and Inscription of the Early History of the Sassanian

Empire, Vols. I-II, (Berlin, 1924). HOFTIJZER, J. and Jongeling, K., Dictionary of the North-West Semitic Inscriptions,

(Leiden, etc., 1995).HUMBACH, H., and Skjærvø, P. O., The Sassanian Inscription of Paikuli, Part 3.1,

Restored Text and Translation and Part 3.2, Commentary, by P. O. Skjærvø,(Wiesbaden, 1983).

LAZARD, G., “Notes de vieux-perse”, BSL, 71, (1976), 175-192.LEVINE, B. A., review of Bowman, Aramaic Ritual Texts, in JAOS, 92, (1972), 70-79.MORANO, E., “Contributi all’interpretazione della bilingue greco-partica dell’Eracle

di Seleucia”, in Gnoli, G., and A. Panaino, (eds.), Proceedings of the FirstEuropean Conference of Iranian Studies held in Turin, September 7th-11th,1987 by the Societas Iranologica Europaea. Part 1: Old and Middle IranianStudies. Part 2: Middle and New Iranian Studies, (Serie Orientale Roma. Vol.67,1. Rome, 1990), 229-238.

NYBERG, H. S., A Manual of Pahalvi, Vol. I, (Wiesbaden, 1964); Vol. II, (Wies-baden, 1974)

NYBERG, H.S., Frahang i pahlavik, ed. B. Utas with the collaboration of C. Toll,(Wiesbaden, 1988)

316 ARAMAIC IN IRAN

Page 35: Aramaic in Iran

Pugliese Carratelli, G., and G. Garbini with foreword by G. Tucci and introd. byU. Scerrato, A Bilingual Graeco-Aramaic Edict by Asoka. The First Greek In-scription Discovered in Afghanistan, (Serie Orientale Roma, XXIX, Rome,1964).

ROSENTHAL, F., “Aramaic. 1. General”, in Yarshater, E., (ed.) Encyclopaedia Iranica,II/3, (London, etc., 1986), 251-256.

SALEMANN, C., “Mittelpersisch”, in Grundriss der iranischen Philologie, ed. W. Geigerand E. Kuhn, Vol. 1/1, (Strassburg, 1895-1901; repr. Berlin and New York,1974), 249-332.

SCHAEDER, H. H., Iranische Beiträge, Vol. I, (Schriften der Königsberger GelehrtenGesellschaft, Geisteswiss. Kl. 6, 5, Halle-Saale, 1930).

SEGERT, S., Altaramäische Grammatik, (Leipzig, 1983). SIMS-WILLIAMS, N., “The Final Paragraph of the Tomb-Inscription of Darius I (DNb,

50-60: the Old Persian Text in the Light of an Aramaic Version”, BSOAS, 44,(1981, 1-7).

SIMS-WILLIAMS, N., “A Sogdian Greeting”, in Emmerick, R. E., and D. Weber(eds.), Papers in Honour of Prof. D. N. MacKenzie on the occasion of his 65thbirthday on April 8th, 1991, (Frankfurt, etc., 1991), 176-187.

SKJÆRVØ, P. O., Undersøkelser til verbalsystemet i gammelpersisk og vestlig middel-iransk (Investigations into the verbal systems of Old Persian and western MiddlePersian), M.A. thesis, Oslo and Copenhagen, 1974.

SKJÆRVØ, P. O., “Case in Inscriptional Middle Persian, Inscriptional Parthian andthe Pahlavi Psalter”, Studia Iranica, 12, (1983), 69-94, 151-181.

SKJÆRVØ, P. O., “Remarks on the Old Persian Verbal System”, Münchener Studienzur Sprachwissenschaft, 45, (1985), 211-227 (= Festgabe für K. Hoffmann).

SKJÆRVØ, P. O., “Verbs in Parthian and Middle Persian Inscriptions”, in Schmitt, R.,and P. O. Skjærvø, eds., Studia Grammatica Iranica. Festschrift für HelmutHumbach, (Munich, 1986), 425-439.

SKJÆRVØ, P. O., “Verbal Ideograms and the Imperfect in Middle Persian and Parthian”,in Études irano-aryennes offertes à Gilbert Lazard, (Studia Iranica, cahier 7,1989), 333-54.

SKJÆRVØ, P. O., review of Nyberg, H. S., Frahang i pahlavik, (Wiesbaden, 1988), inKratylos, 35, (1990), 95-99.

SKJÆRVØ, P. O., “L’inscription d’Abnun et l’imparfait en moyen-perse”, StudiaIranica, 21, (1992), 153-60.

SKJÆRVØ, P. O., review of Gnoli, G., and A. Panaino, (eds.), Proceedings of the FirstEuropean Conference of Iranian Studies held in Turin, September 7th-11th,1987 by the Societas Iranologica Europaea. Part 1: Old and Middle IranianStudies. Part 2: Middle and New Iranian Studies, (Serie Orientale Roma. Vol.67,1. Rome, 1990), in Bulletin of the Asia Institute 8 (1994 [1996]), 319-323.

SKJÆRVØ, P. O., “Aramaic Scripts for Iranian Languages”, in The World’s WritingSystems, (Oxford, 1996), 515-535.

PRODS OKTOR SKJÆRVO 317

Page 36: Aramaic in Iran

SKJÆRVØ, P. O., review of Skalmowski, W., and A. van Tongerloo, (eds.), Me-dioiranica. Proceedings of the International Colloquium Organized by theKatholieke Universiteit Leuven from the 21st to the 23rd of May 1990, (Leuven,1993), in Kratylos, 41, (1996), 107-113.

SKJÆRVØ, P. O., “On the Middle Persian Imperfect”, in the acts of the ColoquioInternacional “Syntaxis de las lenguas indo-iranias antiguas". Sitges (Barcelona)4-6 May 1993, forthcoming.

TOLL, C., “Die aramäischen Ideogramme im Mittelpersischen”, in Diem, W., andA. Falaturi (eds.), XXIV. Deutscher Orientalistentag: vom 26. bis 30. September1988 in Köln. Ausgewählte Vorträge, (Zeitschrift der Deutschen MorgenländischenGesellschaft. Supplement 8, Stuttgart, 1990), 25-45.

UTAS, B., “Verbal Forms and Ideograms in the Middle Persian Inscriptions”, ActaOrientalia, 36, (1974), 83-112.

N. L. WESTERGAARD, Zendavesta or the Religious Books of the Parsis I, (Copenhagen,1852-54, repr. Wiesbaden, 1994).

318 ARAMAIC IN IRAN